This session will dive into the architecture and implementation of a real-time personalization API built on Microsoft Fabric, leveraging the power of Functions, Cosmos DB, Real-Time Intelligence, and Apache Spark. Attendees will learn how to capture and process real-time customer behavior telemetry using Fabric Real-Time Intelligence and land it in an Eventhouse for providing analysts with near real time insights in a BI dashboard.

We will then demonstrate how to build a collaborative filtering-based recommender system using Apache Spark within Fabric’s runtime. This process involves generating user and product vectors whose dot product signifies the propensity of a user to purchase a product. A key focus will be on the efficient reverse ETL of these materialized vectors into Azure Cosmos DB using a blue-green deployment strategy, establishing a high-throughput, low-latency OLTP serving layer.

Finally, the session will showcase how Fabric Functions can host an API that powers a live customer-facing website. This API will perform rapid key-value lookups against Cosmos DB to retrieve user vectors and a matrix of product vectors, enabling the real-time calculation and serving of personalized product recommendations directly to the end-user.

Join us to understand how to build a complete, scalable, and real-time personalization solution on Microsoft Fabric that transforms customer data into engaging and timely experiences.